AI-based Techniques Research Articles

Enhancing Fake Image Detection with a Hybrid Approach of Deep Learning and Image Forensics

With the rapid growth of digital content creation and manipulation tools, verifying the authenticity of visual media has become a growing concern across several domains such as journalism, law enforcement, politics, and social media. The widespread availability of advanced image editing software, coupled with AI-based techniques like Generative Adversarial Networks (GANs), has led to a surge in highly realistic fake images and deepfakes, making traditional methods of detection increasingly ineffective. This research addresses this evolving challenge by proposing a hybrid model that integrates the interpretability of traditional image forensic techniques with the adaptability and learning capacity of Convolutional Neural Networks (CNNs).Traditional forensic methods, such as Photo-Response Non-Uniformity (PRNU), Error Level Analysis (ELA), and lighting inconsistency checks, are well-known for their transparency and ease of interpretation. These approaches typically analyze intrinsic image features like compression artifacts, sensor noise, and inconsistencies in color or geometry. However, they often struggle when detecting content generated by sophisticated AI models, which can produce highly realistic forgeries that bypass classical detection techniques.On the other hand, CNNs have shown significant promise in identifying subtle patterns and anomalies that are often invisible to the human eye and traditional algorithms. These models automatically learn hierarchical features from large datasets, allowing them to generalize across multiple types of manipulations. However, CNNs face challenges in terms of interpretability and are vulnerable to adversarial attacks, which can deceive the model with carefully crafted inputs.This study proposes a hybrid framework that leverages the strengths of both paradigms. By combining CNNs with traditional forensic cues, the model achieves both high accuracy and improved transparency. Experimental evaluations demonstrate the model's robustness across multiple datasets and manipulation types. The results show a significant improvement over standalone methods, confirming the effectiveness of the proposed hybrid approach in detecting fake images. This work contributes to the development of more dependable and explainable systems, suitable for deployment in real-world, high-stakes environments.

Comprehensive optimization of shot peening intensity using a hybrid model with AI-based techniques via Almen tests

Abstract Shot peening is a crucial surface treatment technique that significantly improves the mechanical properties of metallic components, particularly their fatigue resistance and ability to withstand corrosion cracking. This study aims to optimize the shot peening process for aviation applications by evaluating and comparing various mathematical modeling and optimization techniques. Seven mathematical models were analyzed using a neuro-regression method (NRM), among which the second-order trigonometric non-linear (SOTN) model exhibited the highest reliability, achieving R2 values of 0.93 and 0.90 for training and testing datasets, respectively. To improve the model’s robustness, four optimization algorithms – differential evolution (DE), simulated annealing (SA), Nelder–Mead (NM), and random search (RS) – were applied to the SOTN model. Although each technique offered valuable insights, performance fluctuations across different intensity ranges necessitated the development of a hybrid optimization model that combines the strengths of all four methods. The hybrid model achieved a mean error of approximately 2.69 %, outperforming individual approaches and demonstrating strong potential for reliable shot peening optimization across a wide range of target intensities. These findings provide a comprehensive methodology for AI-based optimization of surface treatment processes in engineering applications.

Embedded Artificial Intelligence in Guided Wave SHM system: Signal processing, and data analysis

This article presents an overview of recent advances in the development of guided wave (GW) structural health monitoring (SHM) solutions at CEA List. The proposed approach covers the entire data processing chain, from signal acquisition to automated analysis. Both active and passive acquisition techniques are addressed, and the integration of Fibre Bragg Grating (FBG) sensors for operation in challenging environments is explored. Advanced signal processing and imaging methods are embedded directly into acquisition systems, with a strong emphasis on AI-based techniques. First, an efficient AI-driven method for compensating temperature effects is introduced. In addition, the use of simulations to generate large, annotated datasets is presented as a solution to the scarcity of real defect data, enabling the training of deep learning models for automated diagnosis.

Collaborative approaches to enhancing smart vehicle cybersecurity by AI-driven threat detection

The introduction sets the stage for exploring collaborative approaches to bolstering smart vehicle cybersecurity through AI-driven threat detection. As the automotive industry increasingly adopts connected and automated vehicles (CAVs), the need for robust cybersecurity measures becomes paramount. With the emergence of new vulnerabilities and security requirements, the integration of advanced technologies such as 5G networks, blockchain, and quantum computing presents promising avenues for enhancing CAV cybersecurity. Additionally, the roadmap for cybersecurity in autonomous vehicles emphasizes the importance of efficient intrusion detection systems and AI-based techniques, along with the integration of secure hardware, software stacks, and advanced threat intelligence to address cybersecurity challenges in future autonomous vehicles.

Mapping football tactical behavior and collective dynamics with artificial intelligence: a systematic review.

Football, as a dynamic and complex sport, demands an understanding of tactical behaviors to excel in training and competition. Artificial intelligence (AI) has revolutionized the tactical performance analysis in football, offering unprecedented data analytics insights for players, coaches, and analysts. This systematic review aims to examine and map out the current state of research on AI-based tactical behavior, collective dynamics, and movement patterns in football. A total of 2,548 articles were identified following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines and the Population-Intervention-Comparators-Outcomes framework. By synthesizing findings from 32 studies, this review elucidates the available AI-based techniques to analyze tactical behavior and identify the collective dynamic based on artificial neural networks, deep learning, machine learning, and time-series techniques. Concretely, the tactical behavior was expressed by spatiotemporal tracking data using convolutional neural networks, recurrent neural networks, variational recurrent neural networks, and variational autoencoders, Delaunay method, player rank, hierarchical clustering, logistic regression, XGBoost, random forest classifier, repeated incremental pruning produce error reduction, principal component analysis, and T-distributed stochastic neighbor embedding. Furthermore, collective dynamics and patterns were mapped by graph metrics such as betweenness centrality, eccentricity, efficiency, vulnerability, clustering coefficient, and page rank, expected possession value, pitch control map classifier, computer vision techniques, expected goals, 3D ball trajectories, dangerousity assessment, pass probability model, and total passes attempted. The performance of technical-tactical key indicators was expressed by team possession, team formation, team strategy, team-space control efficiency, determining team formations, coordination patterns, analyzing player interactions, ball trajectories, and pass effectiveness. In conclusion, the AI-based models can effectively reshape the landscape of spatiotemporal tracking data into training and practice routines with real-time decision-making support, performance prediction, match management, tactical-strategic thinking, and training task design. Nevertheless, there are still challenges for the real practical application of AI-based techniques, as well as ethical regulation and the formation of professional profiles that combine sports science, data analytics, computer science, and coaching expertise.

A Systematic Review of AI-Based Techniques for Automated Waste Classification.

Waste classification is a critical step in waste management that is time-consuming and necessitates automation to replace traditional approaches. Recently, machine learning (ML) and deep learning (DL) have gained attention from researchers seeking to automate waste classification by providing alternative computational techniques to address various waste-related challenges. Significant research on waste classification has emerged in recent years, reflecting the growing focus on this domain. This systematic literature review (SLR) explores the role of artificial intelligence (AI), particularly machine learning (ML) and deep learning (DL), in automating waste classification. Using Kitchenham's and PRISMA guidelines, we analyze over 97 studies, categorizing AI-based techniques into ML-based, DL-based, and hybrid models. We further present an in-depth review of over fifteen publicly available waste classification datasets, highlighting key limitations such as dataset imbalance, real-world variability, and standardization issues. Our analysis reveals that deep learning and hybrid approaches dominate the current research landscape, with CNN-based architecture and transfer learning techniques showing particularly promising results. To guide future advancements, this study also proposes a structured roadmap that organizes challenges and opportunities into short-, mid-, and long-term priorities. The roadmap integrates insights on model accuracy, system efficiency, and sustainability goals to support the practical deployment of AI-powered waste classification systems. This work provides researchers with a comprehensive understanding of the state-of-the-art in ML and DL for waste classification and offers insights into areas that remain unexplored.

Artificial Intelligence in Endodontics: Present Uses and Prospective Paths

Artificial intelligence (AI) is a technology that mimics intelligent human behavior by using machines. In recent years, its popularity has grown all over the world. This is primarily due to its capacity to accelerate treatment planning processes, enhance patient outcomes, and improve the accuracy of the diagnosis. To enhance personalized learning, predictive analytics, and patient care plans, endodontic AI-based techniques have been essential in utilizing many models using Deep Learning (DL) and Machine Learning (ML). The purpose of the review was to discuss the current endodontic uses of AI as well as possible future paths. In endodontics, AI models such as (e.g., convolutional neural networks and/or artificial neural networks) are used to study the anatomy of the root canal system, detect periapical lesions and root fractures, determine working length measurements, predict the viability of dental pulp stem cells, and determine the success of retreatment procedures. The future of this technology was discussed in terms of prognostic value diagnostics, drug interactions, scheduling, patient treatment, and robotically assisted endodontic surgery. AI has the potential to be transparent, reproducible, unbiased, and easy to use with careful design and long-term clinical validation. More research is required to verify the cost-effectiveness, applicability, and reliability of AI models before they are routinely used in clinical practice.

Artificial Intelligence in Medicine and Imaging Applications.

Artificial intelligence (AI) can completely transform drug development methods by delivering faster, more accurate, efficient results. However, the effective use of AI requires the accessibility of data of excellent quality, the resolution of ethical dilemmas, and an awareness of the drawbacks of AI-based techniques. Moreover, the application of AI in drug discovery is gaining popularity as an alternative to both the complex and time-consuming process of discovering as well as developing novel medications. Importantly, machine learning (ML) as well as natural language processing, for example, may boost both productivity as well as accuracy by analyzing vast volumes of data. This review article discusses in detail the promise of AI in drug discovery as well as offers insights into various topics such as societal issues related to the application of AI in medicine (e.g., legislation, interpretability and explainability, privacy and anonymity, and ethics and fairness), the importance of AI in the development of drug delivery systems, causability and explainability of AI in medicine, and opportunities and challenges for AI in clinical adoption, threat or opportunity of AI in medical imaging, the missing pieces of AI in medicine, approval of AI and ML-based medical devices.

A mini review on revolutionizing hydrogenation catalysis: unleashing transformative power of artificial intelligence.

The field of hydrogenation catalysis has undergone a revolution due to the application of artificial intelligence (AI) and machine learning (ML), which have opened up new avenues for improving catalyst design, reaction efficiency, and pathway optimization. Trial-and-error techniques are a major component of traditional catalyst discovery methods, and they can be resource and time-intensive when it comes to real-world applications. On the other hand, real-time reaction condition optimization, predictive modelling, and quicker catalyst screening are made possible by AI-based techniques. Using methods like neural networks, Bayesian optimization, and generative models, this paper emphasizes how artificial intelligence has been revolutionizing catalyst creation along with mechanistic knowledge and process intensification. AI has the ability to completely transform catalytic research, as demonstrated by a number of case studies that demonstrate its use in CO₂ hydrogenation, biomass upgrading, and metal catalyzed reactions. This review synthesizes recent developments in AI-enhanced catalytic modelling, kinetic parameter estimation, and multi-scale reaction simulations and explores machine learning models such as Random Forest, Gradient Boosting, Artificial Neural Networks, and Gaussian Processes to predict key catalytic performance indicators. Additionally, high-throughput simulated screening and computational methods such as Density Functional Theory simulations and molecular descriptor-based modelling have been used to improve catalyst design tactics. Summary of the ML models which were trained and validated using open source frameworks such as scikit-learn, TensorFlow, and PyTorch is also presented in this paper. Most of the research studies datasets were using the resource data from Catalysis Hub and the materials project. Techniques for data processing and pre-processing include methods for choosing the component features, such as d-band center analysis, adsorption energy calculations, and algorithm normalization. This review study consists of an in-depth analysis of how data-driven modelling improves catalyst performance, and its prediction and optimization in hydrogenation catalysis reactions by artificial intelligence and machine learning driven approaches.

UMBRELLA REVIEW OF AI-BASED RADIOLOGY TECHNIQUES: COMPARING TRADITIONAL AND DEEP LEARNING METHODS

Background Artificial intelligence (AI) has revolutionized diagnostic radiology by improving accuracy, efficiency, and clinical decision-making. While numerous systematic reviews and meta-analyses have evaluated the performance of AI-based techniques, particularly traditional machine learning (ML) and deep learning (DL) models, a comprehensive synthesis of this evidence is lacking. An umbrella review is necessary to integrate and compare findings across various imaging applications, offering the highest level of evidence to inform clinical practice and policy. Objective This umbrella review aims to compare the diagnostic performance of traditional machine learning and deep learning methods in radiology by synthesizing evidence from existing systematic reviews and meta-analyses. Methods A systematic literature search was conducted in PubMed, Scopus, Web of Science, and Cochrane Library for systematic reviews and meta-analyses published between 2018 and 2024. Only peer-reviewed reviews evaluating diagnostic applications of AI in radiology were included. Methodological quality was assessed using the AMSTAR 2 tool, and risk of bias was evaluated using ROBIS. Results Seven systematic reviews and meta-analyses were included. Across multiple imaging domains—such as breast cancer screening, spinal stenosis, and tumor grading—deep learning consistently outperformed traditional ML in diagnostic accuracy and sensitivity. Evidence quality was rated moderate to high, though variability in reporting and a lack of external validation were noted. Explainability features in DL models were underutilized and poorly evaluated. Conclusion This umbrella review confirms the superior diagnostic performance of deep learning methods over traditional ML in radiology. Future research should prioritize model transparency, real-world validation, and standardization in clinical implementation.

A survey on AI-augmented Secure RTL design for hardware trojan prevention

Once, discrete circuit elements, called components, were heaped up on boards inside steel cages using wire-lead technology in just five short years. Fast forward to today, and your computer CPU fits about half an inch square on a chip. Both this constant miniaturization of electronic circuits and the rapid growth in the prevalence of third-party intellectual property parts have made hardware protection more worrisome than ever. Among all these issues, Hardware Trojans (HTs)—which represent corrupted or harmful additions during various design and fabrication stages—pose significant threats to system integrity, privacy of data, and essential infrastructure. Recent studies have investigated machine learning (ML) and artificial intelligence (AI) techniques designed to enable Hardware Trojans to be found, located, and eliminated in all stages from the register transfer level (RTL) and beyond. This survey gives an in-depth look at how AI can enhance RTL security. It classifies these AI-based techniques into four main categories: Graph-Based Techniques GNNs, for instance, can be used to estimate the topology of circuits, extract structural characteristics, and thus find where some corruption has occurred. The SALTY framework applies Jumping-Knowledge GNNs to improve the accuracy location for hardware Trojans. Deep Learning in Side-Channel and Power-Analysis Techniques Deep learning methods—such as Siamese Neural Networks (SNNs) and Long Short-Term Memory (LSTM) models—have been developed to detect abnormalities brought about by Trojans in power consumption or electromagnetic (EM) radiation, granting non-invasive practices clear security benefits. Studies show that these techniques are superior to the traditional golden-model side-channel detection techniques. Machine Learning Analysis of RTL Code: In conjunction with AI, research teams are now building nearest-neighbor classifiers and decision trees and using reinforcement learning (RL) to recognize occurrences of Trojans inside RTL code. Some research uses Verilog/VHDL conditional statements as features for ML, making it possible for early warning signals to be effectively detected and introducing a proactive security mechanism during the design phase. Comprehensive Security Measures and Logic Locking: A step-by-step methodology has evolved for prevention measures such as logic locking and layout hardening, which aims against a splendid prospect within reach. The TroLLoc framework uses logic obfuscation combined with security-aware placement and routing, thus mitigating security exposures post-design. However, comprehensive studies point out several outstanding problems: key recovery attacks and unintended data leakage related to security in logic locking. In this way, the paper evaluates various AI-driven security strategies in an organized, facilitative manner, thereby highlighting significant challenges and proposing future research directions.

AI-Based Techniques for Classifying Abnormalities Linked to Alzheimer's Disease Progression

Alzheimer's disease (AD) is a brain disorder that gets worse over time and makes diagnosis and treatment much harder. It is very important to find problems early and correctly diagnose them in order to use successful treatment plans. Recently, artificial intelligence (AI), as deep learning (DL) methods, has shown that they could help doctors diagnose and classify problems more accurately in people with Alzheimer's disease. This essay explores how different AI techniques can be used to find and group neuropathological changes that are typical of Alzheimer's disease. Our method uses cutting-edge AI tools like convolutional neural networks (CNNs), recurrent neural networks (RNNs), Residual Networks (ResNet), and MobileNet. These all play important roles in processing and analysing brain images and clinical data. We do a thorough analysis of the current state of AI uses in the imaging and clinical data analysis of AD. We focus on how these models help tell the difference between normal ageing and the early stages of Alzheimer's, as well as how they can be used to stage the disease. Our results clearly show that MobileNet does a better job than other models at classifying problems related to AD. This is because it is better at working with big sets of images. In addition, we talk about how combining different types of data sources makes monitoring tools much more accurate and reliable. It also talks about the problems and moral issues that come up when using AI in hospital settings. This shows how AI has the ability to completely change the way diagnoses are made. AI has the potential to completely change Alzheimer's study because it improves the accuracy of diagnoses and helps make focused treatments..

Artificial Intelligence and Inflation Forecasting: A Contemporary Perspective

The growing complexity of economic systems and the enormous data availability make the application of traditional forecasting methods challenging in accurately predicting economic parameters. A notable shift from econometric models to artificial intelligence (AI) algorithms has significantly affected economic forecasting. This article focuses on the application of AI techniques, specifically in the domain of inflation forecasting. We conduct a comprehensive review by surveying seminal literature on the application of AI in inflation forecasting from the contemporary perspective. This study serves as a pioneering work by consolidating major contributions in the field, offering future researchers’ insights into a diverse array of state-of-the-art AI-based techniques and data sources relevant to inflation forecasting. JEL Classification: E17, E31

Road Detection, Monitoring, and Maintenance Using Remotely Sensed Data

Roads are a form of critical infrastructure, influencing economic growth, mobility, and public safety. However, the management, monitoring, and maintenance of road networks remain a challenge, particularly given limited budgets and the complexity of assessing widespread infrastructure. This Special Issue on “Road Detection, Monitoring, and Maintenance Using Remotely Sensed Data” presents innovative strategies leveraging remote sensing technologies, artificial intelligence (AI), and non-destructive testing (NDT) to optimize road infrastructure assessment. The ten papers published in this issue explore diverse methodologies, including novel deep learning algorithms for road inventory, novel methods for pavement crack detection, AI-enhanced ground-penetrating radar (GPR) imaging for subsurface assessment, high-resolution optical satellite imagery for unpaved road assessment, and aerial orthophotography for road mapping. Collectively, these studies demonstrate the transformative potential of remotely sensed data for improving the efficiency, accuracy, and scalability of road monitoring and maintenance processes. The findings highlight the importance of integrating multi-source remote sensing data with advanced AI-based techniques to develop cost-effective, automated, and scalable solutions for road authorities. As the first edition of this Special Issue, these contributions lay the groundwork for future advancements in remote sensing applications for road network management.

Prediction of the packaging chemical migration into food and water by cutting-edge machine learning techniques

Chemicals transfer from the packaging materials and their dissolution in food and water can create health risks. Due to the costly and time-intensive nature of experimental measurements, employing artificial intelligence (AI) methodologies is beneficial. This research uses five renowned AI-based techniques (namely, long short-term memory, gradient boosting regressor, multi-layer perceptron, Random Forest, and convolutional neural networks) to anticipate chemical migration from packaging materials to the food/water structure, considering variables such as temperature, chemical characteristics, and packaging/food types. The relevance analysis has been employed for monitoring the way that these explanatory variables impact the chemical migration from packaging materials into foods and water. Optimizing the hyperparameters, evaluating the prediction accuracy, and comparing the performance of these AI models reveal that the gradient boosting regressor (GBR) is the superior method for this simulation. The proposed GBR model accurately predicts 1847 experimental datasets, showcasing mean squared error, mean absolute error, root mean squared error, relative absolute error percent, and regressing coefficient, of 0.06, 0.15, 0.24, 6.46%, and 0.9961 respectively. Additionally, implementing a leverage algorithm for outlier detection further affirms the reliability of this modeling study.

Enhancing cardiovascular disease classification in ECG spectrograms by using multi-branch CNN.

Enhancing cardiovascular disease classification in ECG spectrograms by using multi-branch CNN.

Innovations in Wildfire and Smoke Detection: A Comprehensive Survey

Wildfires are a growing threat due to climate change, causing significant damage to ecosystems, property, and human lives. Effective detection systems are critical for prompt intervention and damage mitigation. This survey explores advances in fire and smoke detection, emphasizing the role of machine learning and computer vision techniques such as Convolutional Neural Networks (CNNs) and YOLO object detection models. Recent approaches leverage multi-source data, including satellite imagery, drone feeds, and ground sensors, to enhance detection accuracy and scalability. Key contributions include lightweight models such as FireNet for IoT applications, real-time smoke detection frameworks, and hybrid systems integrating traditional and AI-based techniques. Despite notable progress, challenges remain, such as false positives, environmental variability, and computational limitations in resource-constrained environments. This paper reviews these advancements, evaluates their limitations, and identifies promising research directions for developing robust and scalable wildfire detection systems. Key Words Wildfire Detection, Smoke Detection, Machine Learning, Deep Learning, CNN, YOLO, Disaster Management, IoT, Remote Sensing

AI-based methods for biomolecular structure modeling for Cryo-EM.

AI-based methods for biomolecular structure modeling for Cryo-EM.

Modelling, solution and application of optimization techniques in HRES: From conventional to artificial intelligence

Modelling, solution and application of optimization techniques in HRES: From conventional to artificial intelligence

Forecasting of Fine-Grained SIF of OCO-2 Using Multi-source Data and AI-Based Techniques

Forecasting of Fine-Grained SIF of OCO-2 Using Multi-source Data and AI-Based Techniques